10Oct_1740_Tarbox-Trials.ppt

Download Report

Transcript 10Oct_1740_Tarbox-Trials.ppt

DICOM INTERNATIONAL
CONFERENCE & SEMINAR
Oct 9-11, 2010
Rio de Janeiro, Brazil
Collection and Management of
Image Data for Clinical Trials
Lawrence Tarbox
Washington University in St. Louis
Colin Rhodesa, Steve Mooreb, Ken Clarkb, David Maffittb,
John Perryc, Toni Handzela, Fred Priorb
a
b
c
VirtualScopics Inc., Rochester, NY 14625 USA
Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, MO
Hampshire, IL 60140 USA
Warning!
• Small Type Ahead!
• Do not strain to read!
• I will talk through the
diagrams.
• Slides will be available
for download to see the
details
Imaging Biomarkers in
Clinical Trials
• Quantitative image analysis requires digital image
data, and often non-standard (i.e. non-clinical) image
acquisition parameters.
 To access sufficiently large participant
populations, clinical trials are usually
multi-center but increasingly they are
also international.
 Research imaging protocols frequently
generate more data than clinical studies.
All of this digital data must be
acquired, managed, processed, and
ultimately reduced to meaningful
scientific evidence presented to a
sponsor.
Efficiency and Quality
Are Critical
Timing Histogram

60%
Percentage
50%
40%

30%
20%
10%
145
135
125
115
105
95
85
75
65
55
45
35
25
15
5
0%
55% of Imaging Studies
were received within 5
days, 86% received
within 20 days.
Anecdotal Industry
information points to
5%-50% image fallout
rate heavily dependant
on acquisition modality.
Days
Time from scan date to receipt at the core lab
Data Courtesy of Jeff Markin
NCI 1R41CA132790-01 (PI: Rhodes, Prior)
Using an Electronic Imaging Trial Management
System Transmission Delays
and Data Loss are Reduced
• (68.7%) are
received within
5 days;
• 95% within 20
Days;
• Data Drop-out
Rate < 5%.
Time from scan date to receipt at the core lab
B. Vendt, R. McKinstry, W. Ball, M. Kraut, F. Prior, and M. DeBaun, “Silent Cerebral Infarct Transfusion (SIT) Trial
Imaging Core: Application of Novel Imaging Information Technology for Rapid and Central Review of MRI of the Brain,”
Journal of Digital Imaging, vol. 22, pp. 326-343, 2009.
Regulatory Compliance
• Data submissions to the FDA in support of
pharmaceutical trials require documentation of all
access to the data from its time of creation to the point of
submission (Provenance Tracking, Digital Signatures).
– Any modification of the data (including image processing) must be
documented and electronically signed.
• Image visualization systems (PACS) are FDA regulated
medical devices and therefore subject to strict
regulations on software development and validation.
• Software that manages data collected during clinical
trials is also subject to FDA regulation (GMP).
• Open source software is permitted but only after
carefully controlled validation and verification.
Regulatory Compliance
• Software Process: Code of Federal Regulations, Title 21
parts 820, 2009. Available from:
http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSe
arch.cfm?CFRPart=820&showFR=1
• Provenance Tracking: Code of Federal Regulations,
Title 21, part 11, 2009. Available from:
http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfCFR/CFRS
earch.cfm?CFRPart=11&showFR=1
Software can NOT comply with these regulations.
The regulations relate to processes that may involve
software, not the software itself.
Three Main Components
• Located at each study site
– Clinical Studies Anonymization Workstation
(CSAW)
• Located at the Contract Research
Organization (CRO)
– Image Check-in and Quality Assurance
Management System (ICMS)
– Information Repository Management System
(IRMS)
CSAW Implementation
• Principally based on the Clinical Trials
Processor (CTP), part of RSNA MIRC
– Configurable pipeline for receiving, de-identifying,
checking, processing, and transmitting data
• Creates IHE TCE Manifest
• Digital Signatures as integrity checks
• Includes additional web-based management
components
– Out-of-band confirmation of transmission
– Collection of data missing from the DICOM objects
Check-in and Quality Check
• Check received images against
manifest
• Automated check for required
Attributes
• Quarantine of images that do not pass
checks
• Optional manual check for image
quality
• Release to IRMS for further processing
Imaging Trial Data Analysis
• Reader studies:
– Multiple blind reads with or without consensus;
– Radiologist generated measurements;
– Electronic Case Report Forms linked to the imaging study
should facilitate workflow and reduce transcription error.
• Quantitative analysis:
– Semi-automated - human guided analysis software;
– Automated imaging pipelines.
• Image management and distribution are key to all
forms of analysis.
• Workflow management and optimization are
increasingly important as data volumes grow.
IRMS Prototype
:
caBIG-IMG-xx-01-05112007 (Azar: PI)
caBIG-IMG-82-01-04102008 (Schwanke: PI)
Validation
• A feasibility study is underway linking one
imaging site (Washington University) and one
CRO imaging core lab (VirtualScopics).
1) the elapsed time from image capture to when the images are measured
or reviewed at the central processing center; and
2) the number of imaging protocol and FDA compliance errors found at
each step in the workflow from image capture through analysis at the
central processing center
• Audit of Development Process documentation
and controls.
• Audit of Trial Processes and Procedures for
Part 11 compliance based on electronic
Provenance tracking.
In Summary
• Quantitative digital imaging is playing an
increasingly important role in clinical trials.
• Collecting, managing and analyzing images in
clinical trials require specialized software
systems.
• Management development processes must
meet FDA quality requirements and the
software itself must support Part 11 compliant
processes.
• Once such management systems are deployed,
sponsors have the opportunity for improved
trial monitoring and information access.
Acknowledgements
This project funded in part by the National
Institutes of Health through an R41 SBIR grant
CA132790-01A1
The CTP project is funded through RSNA
The XIP and AVT projects are partially funded
through numerous subcontracts funded by
National Cancer Institute